Cairn (Product): news timeline & CVEs

AI tools help hacker break in for $25 per target

AI-driven campaign compromised 27 of 105 retailers and stole about 600,000 card numbers.

Israeli security firm Gambit reports that an attacker used open-source AI harnesses to hit 105 online retailers over five days and compromise 27. From two businesses the attacker took about 600,000 active credit card details, and card-skimmer scripts were installed at five more, with additional access to unnamed major companies. Gambit identifies Strix for vulnerability search, Cairn for autonomous exploitation, and Hermes for campaign orchestration, with model access via OpenRouter. An August 25 balance showed $7,005 spent over four weeks, about $25 per target, ranging from $3.13 to $79.31.

CSO OnlineSep 25, 17:38 UTC · 1d agoData breach in the wild 6 sources

Crook used three open source agents to break into a Fortune 500 hospitality company, a major US airline and 25+ other orgs

An operator used three open-source AI agents to compromise at least 27 organizations and steal over 600,000 card records.

Security firm Gambit reconstructed a campaign in which a Chinese-speaking operator used the open-source AI harnesses Hermes, Strix, and Cairn to attack retailers and other companies. Between September 10 and 15 the operator launched at least 105 attacks and compromised at least 27 organizations, including a Fortune 500 hospitality company and a major US airline. Two victims lost more than 600,000 credit card records, and skimmer scripts were confirmed on 19 websites, with over 100 additional infected sites reported. One intrusion used SQL injection, a web shell, a misconfigured sudo rule, and theft of AWS credentials and 46 secrets; Gambit estimates total model costs at $12,000 to $18,000.

The Register · Securityupdated Sep 25, 17:38 UTC · 1d agofirst Sep 24, 23:32 UTC · 2d agoData breach in the wild 6 sources

AI-Powered Campaign Targets Hundreds of Online Retailers

A Chinese-speaking actor used autonomous AI agents to compromise online retailers and steal over 600,000 credit cards.

Gambit says a Chinese-speaking, financially motivated actor has targeted hundreds of online retailers since July 2026 using three open-source AI agent harnesses. Between 10 and 15 September the actor launched 105 attack projects and compromised at least 27 companies, stealing more than 600,000 unexpired credit cards from two victims, including over 488,000 US cards, and injecting checkout skimmers. Strix on GLM 5.2 and DeepSeek v4 Pro hunted flaws, Cairn on DeepSeek v4.1 Flash drove exploitation, and Hermes on Anthropic opus-4.6 handled orchestration from short Chinese prompts. With researcher Varys, Gambit linked the skimmer to more than 100 additional sites and estimated a mean cost of $25.46 over 101 completed scans.

SecurityWeekupdated Sep 25, 17:38 UTC · 1d agofirst Sep 24, 12:48 UTC · 2d agoThreat actor in the wild 6 sources

Malicious AI agents steal 600K credit cards, infect 100+ sites with skimmers

An AI-agent campaign stole over 600,000 cards and skimmer-infected at least 119 retail sites.

Gambit reports a financially motivated operator, assessed as Chinese, has used open-source AI agent tools since at least July to compromise online retailers, with activity ongoing as of September 22. Strix performed vulnerability scanning, Cairn sought shells or admin access, and Hermes orchestrated post-exploitation with claude-opus-4.6. The actor stole more than 600,000 valid card records from two companies, planted skimmers on at least 119 sites, and hit organizations including a Fortune 500 hospitality firm and a major U.S. airline. Cleanup instructions that wiped Magento card fields after theft also caused data loss, while estimated model spend was about $12,000 to $18,000, roughly $25 per target.

BleepingComputerupdated Sep 25, 17:38 UTC · 1d agofirst Sep 23, 16:20 UTC · 3d agoThreat actor in the wild 6 sources

This Windows Malware is Built to Let Up to Four AI Models Vote on Its Next Move

Talos analyzed CLOSEDQUORUM, Windows malware that lets up to four AI models vote on its next action.

Cisco Talos describes CLOSEDQUORUM, Windows malware that asks up to four AI models—DeepSeek, Qwen, Mistral, and Gemini—to vote on steal, inject, persist, or move, then runs the majority choice. Steal dumps LSASS, copies Chrome, Edge, and Firefox passwords, and collects MetaMask, Exodus, and Ethereum wallet data, sending results through a Discord webhook. Talos has not seen the full chain work, and public builds contain placeholder API keys. The code dates to at least June 17, 2026; Talos published Snort rule 1:66984, YARA logic, six SHA-256 hashes, and the CAIRN hunting tool.

The Hacker Newsupdated Sep 24, 05:44 UTC · 3d agofirst Sep 23, 14:17 UTC · 3d agoMalware 13 sources

Cisco Talos Launches CAIRN Tool to Hunt and Track AI-Integrated Malware

Cisco Talos released CAIRN, an open-source toolkit to identify AI-integrated malware by analyzing metadata for 'cognitive artifacts' from LLMs.

Cisco Talos has launched CAIRN, an open-source research toolkit designed to help defenders hunt and track malware that integrates AI and large language models. The tool uses a metadata-first approach to identify 'cognitive artifacts' like prompt templates and API endpoints left by developers. CAIRN analyzes data from sources like VirusTotal without executing binaries, using filters to flag AI provider domains and frameworks. It was released with an analysis of CLOSEDQUORUM, a malware sample that uses commercial LLMs for autonomous C2 decisions.

GBHackersupdated Sep 24, 05:44 UTC · 3d agofirst Sep 23, 07:39 UTC · 4d agoTools 13 sources

A New Tool for Tracking AI Malware Uncovers Malware That Operates Without Human Control

Cisco Talos releases CAIRN to track AI malware and documents CLOSEDQUORUM, an unconfirmed Windows implant steered by LLMs.

Cisco Talos released CAIRN, an open-source toolkit that hunts AI-integrated malware by scoring prompt templates, provider endpoints, API-key prefixes, libraries, and sandbox metadata without downloading or executing binaries. Its launch describes CLOSEDQUORUM, called the first publicly documented Windows implant to delegate tactical decisions to a panel of commercial models: DeepSeek, Qwen, Mistral, and Google Gemini. Static analysis shows prompts that can select LSASS dumping, browser-password and cryptocurrency-wallet theft, process hollowing, APC injection, persistence, and encrypted exfiltration through Discord. Talos has not confirmed real-world use, and the public build is nonfunctional.

Cyber Security Newsupdated Sep 24, 05:44 UTC · 3d agofirst Sep 22, 15:27 UTC · 4d agoMalware 13 sources

Researchers uncover malware that uses AI to choose its next move

Cisco Talos documents CLOSEDQUORUM, a Windows implant that lets commercial LLMs vote on its next action.

Cisco Talos released CAIRN, an open-source framework that hunts AI-integrated malware using file metadata such as prompts, provider endpoints, and sandbox-evasion text, without executing samples. Its first highlighted find is CLOSEDQUORUM, a 16.4 MB Go Windows implant that asks DeepSeek, Qwen, Mistral, and Gemini to vote on stealing data, injecting code, or establishing persistence, with the goal of harvesting credentials and crypto wallets. The examined copy contained placeholder API keys and a dummy webhook, so researchers did not observe a complete live operation. Talos linked the broader hunt to LAMEHUG, which CERT-UA reported in July 2025 as an early AI-integrated malware sample.

Help Net Securityupdated Sep 24, 05:44 UTC · 3d agofirst Sep 22, 13:56 UTC · 4d agoMalware 13 sources1

Cairn

Timeline8 articles

Appears with